My current setup:
Fitbit Charge 4 for continuous HR and activity data. HR not accurate during activity. Even without activity HR is not perfect, but acceptable and enough for monitoring (for example early disease detection). SPO2 seems to be inaccurate and not useful. Sleep stages inaccurate. Time in bed / Total sleep time seems to be accurate enough to see trends. Not sure on HRV, but sometimes it differs from Oura (which have good science and validation behind HR / HRV). API is available for detailed intraday HR and Activity data.
Oura ring for nightly HR, HRV and temperature deviation data. Sleep stages inaccurate (compared with Dreem 2 ECG during 137 nights). Time in bed / Total sleep time seems to be accurate enough to see trends (but differs from other trackers). Ring shows too much steps (because of hand movements). API is available and simple with 5 min resolution for HR / HRV.
Wellue SPO2 ring for nightly HR and SPO2. Easy export to CSV.
Polar H10 for training HR tracking (most accurate, good science behind). Its not a fitness tracker, but there is no accurate wristband for accurate tracking HR during a training (maybe Apple Watch can do the job).
HR beetween rings and fitbit correlate around r=0.9 for nightly data which is a good sign.
Stopped using:
Garmin Vivosmart 4 - HR was less accurate compared to Fitbit and Polar H10. SPO2 inaccurate (seems true for all wristbands). Stress metrics doesnt make sense because based on inaccurate HR
Polar Vantage V2 - HR isnt accurate enough.
Withings GO - steps very accurate, no HR data. Old device. Can be used as a clip (wrist free).
MetaMotionR - no HR, but have some interesting sensors (for example ambient light). Very buggy software, but have export to CSV. Not suitable for casual use.
You can look there for trackers accuracy comparison - https://www.youtube.com/c/TheQuantifiedScientist/videos